You finish an article and remember the argument, the tone, maybe one excellent sentence. The five unfamiliar words you looked up along the way are usually gone a few days later.
That is not a reason to stop reading difficult material. It is a workflow problem.
A word encountered in a real article has several advantages over a word pulled from a generic list: you chose the text, you cared enough to notice the word, and the surrounding sentence already gives it meaning. The challenge is preserving that context long enough for repeated encounters to turn recognition into recall.
Here is a simple system for doing that.
1. Save selectively, at the moment of reading
Do not save every unknown word.
Save a word when at least one of these is true:
- it blocked your understanding of the sentence;
- you have seen it before and still could not recall it;
- it feels useful for the kinds of things you read or write;
- the phrase around it is worth remembering as a whole.
This matters because a vocabulary library can become a second unread inbox. Ten useful words you review are worth more than a hundred words you captured because the interface made it possible.
The ideal capture action is short enough that you can return to the sentence immediately. If saving requires opening another app, creating a card, finding a definition and reconstructing the source sentence, the friction compounds quickly.
2. Keep the sentence, not only the definition
A dictionary definition is useful for checking meaning. It is a weak memory cue by itself.
Compare these two review cards:
isolated
meticulous — very careful and precise
in context
“She kept meticulous records of every change to the manuscript.”
The second version gives your memory more to work with: syntax, topic, tone and the exact moment where the word made sense.
When you save vocabulary from reading, keep at least:
- the word or expression;
- the original sentence;
- a concise meaning;
- the source title or URL;
- pronunciation when it is useful for the language.
That is enough context to reconnect the review with the original reading experience without turning every card into a research project.
3. Try to recall before revealing the answer
Recognition feels easier than recall. That is exactly why rereading a list of saved words can create false confidence.
During review, look at the word in its sentence and try to retrieve the meaning before revealing it. You do not need a perfect dictionary definition. If you can explain the meaning in your own words and understand the sentence, that is already a useful retrieval attempt.
For productive vocabulary, you can go one step further: hide the target word and try to reconstruct it from the sentence. But do this only for words you genuinely want to use, not every word you want to recognize while reading.
4. Space reviews instead of cramming
A saved word does not need to appear every day forever. The useful interval depends on how stable the memory already is.
Spaced-repetition systems schedule the next review farther into the future when recall is easy and bring difficult items back sooner. Modern systems such as FSRS model this at the card level rather than using one fixed sequence for everything.
The exact algorithm matters less than one practical rule: review a small due queue instead of repeatedly rereading the entire vocabulary list.
Sipword uses an FSRS-based scheduler for this loop. Anki also supports FSRS if you prefer a mature general-purpose flashcard system.
5. Let repeated reading do part of the work
Not every saved word needs to graduate through flashcards alone.
When a word matters in the material you read, you are likely to meet it again. That second or third natural encounter is valuable because it tests whether the memory transfers beyond the original sentence.
This is one reason vocabulary from your own reading can be more useful than an arbitrary deck: your reading habits act as a filter. If you keep encountering a term in economics essays, technical documentation or literary criticism, the word is probably relevant to your language use.
A low-friction weekly routine
A sustainable version can be small:
- Read normally.
- Save only the words worth another encounter.
- Keep the original sentence automatically whenever possible.
- Review what is due for a few minutes.
- Delete words that turn out not to matter.
Deleting is part of the system. A vocabulary collection is not an archive that must preserve every lookup. It is a working set of words you want to make easier to recognize or use.
What to use today
If you want maximum control, Anki is a strong option, especially if you already have a reliable way to create cards from your browser.
If you already read and highlight inside a read-it-later system, Readwise Reader can keep your source material and annotations together.
Sipword is being built specifically around the missing middle: save a word from the page with its context, then review it later without turning reading into card maintenance.
For the broader philosophy behind that workflow, read The case for reading slowly in a second language. For tool tradeoffs, see Anki vs Readwise vs Sipword for vocabulary from reading.